#221 2004 · MediaTek · Semiconductor / mobile phone chipsets

Launching a phone brand used to require a $100 million R&D budget and two years; MediaTek packaged the entire chip-to-firmware stack into one kit and cut that down to a few million yuan and four months.

the problem

a huge upfront R&D investment locks an entire industry to a small number of large players

background

Building a mobile phone in the early 2000s required a manufacturer to design or license a baseband chipset, write and debug its firmware, and build a reference hardware platform from scratch, a process that typically cost upward of $100 million and took 18 to 24 months even for an established electronics company. That barrier limited the phone market to a small number of well-capitalized brands and their approved manufacturing partners.

The obvious business for a chipmaker is to sell chips and license reference designs to established manufacturers who still do the integration work themselves — a model that keeps the chipmaker's addressable market to the handful of companies capable of doing that integration. It leaves untouched a much larger population of small workshops and traders who could sell phones if the technical barrier to actually building one were lower.

what everyone would do

Keep selling chips and reference designs to established manufacturers who do their own integration — the standard chipmaker business model. It keeps the chipmaker's own addressable market limited to the small number of companies with the capital and engineering capacity to integrate a chip into a working phone, leaving a much larger population of smaller workshops locked out not by lack of demand, but purely by the R&D barrier.

what they saw

MediaTek saw that the R&D barrier to making a phone wasn't a natural feature of the market, it was an artifact of how chips had always been sold — as raw components requiring the buyer to do the integration work themselves. Package the entire stack, chip, firmware, reference hardware and testing, into one ready-to-use kit, and the integration work disappears as a barrier, expanding the market of potential phone-makers from a handful of large companies to any workshop that can assemble a case around a working kit.

the move

Starting around 2004, Taiwan's MediaTek sold a fully integrated 'turnkey' solution: a single chip bundled with pre-written firmware, reference hardware designs and testing support, so a manufacturer needed only to choose features, design a case, and assemble the phone. A workshop with essentially no in-house engineers could go from an empty room to a shippable phone in as little as four months on a budget of a few million yuan, sourcing components like cases and cameras separately and letting MediaTek's kit handle everything electronic underneath.

why it works

The roughly $100 million, two-year barrier to making a phone came almost entirely from designing and debugging chip-firmware-hardware integration, not from anything specific to assembling the finished device. Bundling a pre-integrated, pre-tested turnkey solution moved that integration cost into MediaTek's own upfront product development, amortized across every customer who bought the kit, converting phone-making from an engineering problem only well-capitalized companies could solve into an assembly and design problem a small workshop could handle in months on a tiny budget. Because MediaTek could now sell the same kit to a vastly larger population of buyers than the handful of manufacturers previously capable of integration, its own volume multiplied even as each individual sale was lower-value than a full licensing deal with a major brand — chipset volume grew roughly a hundredfold in five years as the resulting shanzhai industry flooded the market with phones.

the payoff

MediaTek's China chipset volume grew from about 3 million units in 2004 toward a projected 300 million by 2009, roughly a hundredfold increase in five years, and the company came to control around half of China's feature-phone chipset market with over 60 percent gross margins on roughly $1.6 billion in annual handset-chip revenue by 2009. The resulting 'shanzhai' phone industry, centered on Shenzhen's Huaqiangbei district, produced an estimated 145 million phones in 2009 alone, close to 30 percent of China's total domestic phone market that year, sold for well under $100 each with features like dual SIM slots that established brands didn't offer.

where it breaks

The strategy only works while the barrier being collapsed is the sole meaningful obstacle between a buyer and market entry — once a different barrier emerges that the turnkey kit doesn't address, the model stops working, which is exactly what happened when the industry shifted to 3G smartphones and software complexity became the bottleneck instead of baseband chip integration, and MediaTek's China handset-chip revenue fell nearly 40% as a result. It also depends on the supplier being willing to enable a flood of smaller, less sophisticated competitors into the market, which expands the supplier's addressable market but can commoditize the end product and create quality-control or reputational issues across the wider industry. And it requires the packaged solution to genuinely be good enough for buyers' real needs — a turnkey kit too limited or generic wouldn't satisfy actual market demand even if it lowered the barrier to entry.

what came after

MediaTek's turnkey model is studied as a Harvard Business School case ('Shanzhai! MediaTek and the White Box Handset Market') and cited in innovation-economics literature on 'good-enough innovation' as a clear example of an enabling-technology supplier collapsing an entire industry's barrier to entry from an R&D problem into an assembly problem. The shift also proved temporary: MediaTek's own China handset-chip revenue fell nearly 40 percent year over year in early 2011 once the transition to 3G smartphone operating systems moved the barrier back toward software MediaTek didn't yet control.